From economist-agents
Define the writing standard for every article in the content pipeline. Use when configuring the Writer Agent, when reviewing article prose quality, when tuning the deterministic polish stage.
npx claudepluginhub oviney/economist-agentsThis skill uses the workspace's default tool permissions.
The definitive reference for the Stage 3 Writer Agent, Stage 4 Editorial Reviewer, and Article Evaluator. Every article must meet these standards before publication. The gold standard is editorial prose combining data-driven reporting with sharp, opinionated analysis in a confident, conversational tone.
Applies Acme Corporation brand guidelines including colors, fonts, layouts, and messaging to generated PowerPoint, Excel, and PDF documents.
Guides strict Test-Driven Development (TDD): write failing tests first for features, bugfixes, refactors before any production code. Enforces red-green-refactor cycle.
Share bugs, ideas, or general feedback.
The definitive reference for the Stage 3 Writer Agent, Stage 4 Editorial Reviewer, and Article Evaluator. Every article must meet these standards before publication. The gold standard is editorial prose combining data-driven reporting with sharp, opinionated analysis in a confident, conversational tone.
research-sourcingeditorial-illustrationarticle-evaluationdefect-preventionFirst sentence must make the reader stop — specific fact, surprising statistic, or provocative assertion.
Banned openings: "In today's world", "It's no secret", "The arrival/emergence/rise of", "When it comes to", "Amidst", any sentence starting with "The" + abstract noun.
Every article must have a specific, debatable argument stated in the first two paragraphs. "AI tools overpromise" is a topic. "AI test generators make maintenance worse because they optimise for coverage metrics that don't correlate with quality" is a thesis.
Lists belong in infographics, not editorial writing. Each point argued in its own paragraph with transitions.
Banned: "it would be misguided", "one might", "it is worth noting", "it should be noted", "it is important to", "further complicating matters", "invites closer scrutiny", "in practical terms".
At least 2 named companies or individuals with specific anecdotes. Banned generic attributions: "organisations", "professionals", "studies show", "experts say", "research indicates".
Remove: throat-clearing ("It goes without saying"), redundant attribution ("As mentioned earlier"), weak intensifiers ("very", "quite", "rather", "fairly").
Banned closings: "will rest on" / "depends on" / "the key is", "In conclusion" / "To summarise", "Only time will tell", "remains to be seen".
One heading per 250-350 words. Headings should be noun phrases that advance the argument, not questions or descriptions.
Make the claim first, attribute second. Never start a sentence with a source name + reporting verb.
Use a colon to add a surprising twist. Banned: starting with "Why"/"How", starting with "The Impact/Role of", longer than 10 words without a twist.
The Economist voice is: confident (states opinions as observations), witty (dry humour, understated irony), British (organisation, analyse, colour), active ("Companies are racing" not "it is being observed"), conversational (brilliant dinner companion, not textbook), precise (every word chosen deliberately).
| Rationalization | Reality |
|---|---|
| "Lists make content more scannable" | Scannable is for blog posts; editorial prose argues through connected paragraphs |
| "We need to hedge to be accurate" | Hedging signals uncertainty; The Economist states opinions as observations and lets the evidence do the qualifying |
| "The opening needs context before the hook" | Context is throat-clearing; the hook IS the context — a striking fact frames the whole piece |
| "Longer articles are more thorough" | Density beats length; a tight 700-word argument outperforms a padded 1200-word survey |
| "Generic attribution is safer" | It's lazier — "Microsoft's testing team found" is both more credible and more engaging than "organisations report" |
| "Summary endings are expected" | Expected by whom? The reader already knows the thesis; end with a provocation that extends the argument |
^[-*\d+\.] returns no matches outside Referenceswc -w on body after frontmatter## count ≤4